Description of problem: Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Current deployment via separate set of tools for Ceph and OSP causes chicken and egg dependency problem. Ceph need OS with all packages that Ceph requires to be deployed on all nodes where Ceph to be deployed with Ceph MON service on a separate nodes from OSD ones. In addition, Ceph requires its cluster network to be setup for OSD/storage nodes and network that ceph client will use that is also OSP storage network.
*** Bug 1119881 has been marked as a duplicate of this bug. ***
Fixed in ceph-deploy 1.5.19.
I think we're now converging on a single Puppet-based installer workflow for OSD and MON nodes using Red Hat Ceph Storage v1.3/Stockwell for OSP7. Suggest the target release is updated accordingly. Adding Federico to CC who is the product manager focused on the Ceph installer component.
Mike, We've now completed the ceph integration with OSP-d. Is this RFE still relevant? If the feature has been satisfied, we'll close this RFE. please advise.
I'd suggest we consider this addressed, but I would also like to ask Arkady and others to feel free to open additional bugs to track features or configuration points that are not fully covered.
Arkady, can we close this? We are tracking more granular features for Ceph in individual BZs now.
Neil, I do not think so. There are 2 issues. 1. There are some upstream pull requests for support of RGW on controller nodes. - But I do not see BZ for it. If one exist, please, cc me on it. 2. Current OSP-D does not support Calamari. I had not seen BZ for it either. Once there are BZs for them epoch BZ can be closed. We will need a separate continues stream of BZs for OSP-D (and heat templates for it) to expose new functionality per Ceph release that should be exposed for OpenStack, especially if default config setup should be changed. Thanks, Arkady